Comparison of Data Mining Classification Methods for Predicting Credit Appropriation through Naïve Bayes and Decision Tree Methods

Abstract

The problem statement of this study was seen on inaccurate assessment of the debtors’ ability in paying off the loan of their businesses so that it often caused credit problems. Data Mining was used in assessing or predicting creditworthiness for a prospective debtor. The author attempted to compare the data mining classification to analyze the credit feasibility prediction through Naïve Bayes and Decision Tree methods. The data of the prospective debtors had been processed through the stages of data mining – Naïve Bayes and Decision Tree. The data were tested through k-folds cross-validation (k = 10). The result of this study was that the accuracy of the method of Decision Tree (J-48) was higher than that of the method of Naïve Bayes. The result of the comparison of the two algorithms was that the Decision Tree (J-48) algorithm had an accuracy of 95.24% and the Naïve Bayes algorithm had an accuracy of 79, 59%.

Keywords: Credit, Naïve Bayes, Decision Tree, K-Folds Validation